1724
Organometallics 1983, 2, 1724-1730
Reactions of Dichlorobis(p-chloro)bis( pentamethylcyclopentadienyl)dirhodium and -diiridium with Hexamethyldialuminum Amelio Vdzquez de Miguel, Manuel Gbmez, Kiyoshi Isobe, Brian F. Taylor, Brian E. Mann, and Peter M. Maitilis” Department of Chemistry, The University, Sheffield 5 3 7HF, England Received June 16, 1983
Reaction of [(C5Me5M)2C14] (M = Rh or Ir) with AlzMe6gives a yellow solution that has been analyzed by low-temperature 13CNMR spectroscopy at 25 and 100 MHz and has been shown to contain Al2Me4Cl2 for which structures 3 are suggested. and heterotrimetallic species formulated as [ (C5Me5MMe3)zA1Me], Reaction of 3a (M = Rh) with acetone gives the trans-bis(pmethy1ene) complex 2, and a mechanism for this transformation is proposed. The complexes 3a and 3b react with other ligands L [dimethyl sulfoxide (Me,SO), triphenylphosphine, and bis(dipheny1phosphino)methane (dppm)] to give the complexes [C5h”e5MMe2L].
Introduction We have recently described the formation of the novel complex dimethylbis(p-methylene)bis(pentamethylcyclopentadieny1)dirhodium (2) from the reaction of dichlorobis(p-chloro)bis(pentamethylcyclopentadieny1)dirhodium (la) with hexamethyldialuminum followed by reaction with a n oxidizer (air) or a hydrogen acceptor (acetone).lv2 Similar but more complex reactions take place when the iridium complex l b is reacted with hexamethyldialumin~rn.l~,~ We have also briefly reported that, on reaction of l a or l b with A12Me6in hydrocarbons a t low temperature and before workup or addition of other reagents, only one rhodium- (or iridium-) containing species is present, as shown by detailed 13C NMR studies. This species was originally formulated as A based on results obtained a t -90
A
“C and a 13C frequency of 25 MHz. The data have now been supplemented by 13C studies a t 100 MHz which, together with the improved integration facilities, have caused us to make a change in some details of the proposed structure. These studies form the subject of this paper.
Results and Discussion Low-Temperature NMR Spectra. The ‘H NMR spectrum of the yellow solution formed by reaction of a solution of A12Me6 with a suspension of l a in perdeuteriotoluene (C7D8)was not very informative, even a t low temperature, owing to the overlap of signals in the metal-methyl region. However, the single C5Me5 resonance a t 6 1.34 (at -90 “C) suggested the presence of only one rhodium complex. The 13C(lH)spectra in C7D8(or cyclopentane) at -90 “C were simple (Figure 1 j and indicated that only one rhodium-containing species (3a) was present. This was characterized by sharp resonances a t 6 -17.98 [d, J(Rh-C) (1) (a) Isobe, K.; Bailey, P. M.; Maitlis, P. M. J. Chem. SOC.,Chem.
(b)Vizquez de Miguel. A.; Isobe, K.; Taylor, B. F.; Nutton, A.; Maitlis, P. M. Ibid. 1982, 758. (2) Isobe, K.; VBzquez de Miguel, A.; Bailey, P. M.; Maitlis, P. M. J. Chem. SOC.,Dalton Trans. 1983, 1441. (3) (a) Isobe, K.; Andrews, D. G.; Mann, B. E.; Maitlis, P. M. J. Chem. SOC., Chem. Commun. 1981, 809. (b) Iaobe, K.; Maitlis, P. M. J . Organomet. Chem. 1983,250, C25. Commun. 1981,809.
0276-7333/83/2302-1724$01.50/0
= 28.1 Hz], +2.61 [d, J(Rh-C) = 23.3Hz], 8.52 (s), and 97.89 [d, J(Rh-C) = 2.7 Hz]. Off-resonance decoupling showed the signals at 6 -17.98, 2.61, and 8.52 to arise from methyl groups; the signal at 6 97.89 came from the carbons of a single C5Me5ring. Assuming similar NOE’s for the three methyl resonances, the relative intensities of the signals, 2:1:5, indicate a “C5Me5RhMe3”unit in which one of the rhodium-bound methyls is different from the other two. These relative intensities remained constant over a large number of integrations involving some 20 different experiments carried out on two different instruments; pulse widths and repetition times were chosen to ensure that the spectra were measured under fully equilibrated conditions in order to optimize accuracy of integration. The msgnitudes of the coupling constants are consistent with these methyls being bound to rhodium4 (see also values for 4a, 5a, and 6a in Table I) and the splitting into doublets shows that they are only associated with one rhodium. Four other resonances may be distinguished in the metal-methyl region. Two of these (6 -4.54 and -7.19) are due to the bridging and terminal methyls respectively of A12Me6,which is present in excess. The other two resonances (6 -5.03 and -5.89) are assigned to other Me-A1 species. The 6 -5.03 resonance is sufficiently close to that for the four equivalent terminal methyls of Al2Me4Cl2 (normally a t 6 -5.12 under similar conditions5p6)for us to identify it with the presence of this molecule. This leaves only one resonance, a t 6 -5.89, to be accounted for. The relative intensities also of these resonances (other than those of A1,Me6) remained quite constant from one experiment to another and on varying the molar ratio of la: A1,Me6 from 2.5 to 1O:l. Integration shows the 6 -5.03 and -5.89 resonances to be in the ratio of 4:0.5 taking the Rh-Me a t 6 2.61 as unity. This indicates an empirical formula for the species of (C5Me5RhMe2Me),A1Me(3a) plus 2(A12Me4C1,). Above about -60 “C the three resonances due to Al2Me6 and A12Me4C12begin to broaden and coalesce. The small resonance a t 6 -5.89 appears not to participate in the exchange a t this temperature and remains separate and reasonably sharp up to ca. -30 “C. Spin saturation transfer (4) Mann, B. E.; Taylor, B. F. =13CNMR Data for Organometallic Compounds”; Academic Press: New York, 1981; p 44. ( 5 ) G6mez. M.; Spencer, C. M.; Maitlis, P. M., unpublished results. (6) (a) For discussions of organoaluminum chemistry, see: Mole, T.; Jeffery, E. A. ‘Organoaluminum Compounds”;Elsevier: New York, 1972. (b) Eisch, J. J. ”Comprehensive Organometallic Chemistry”; Wilkinson, G., Stone, F. G. A., Abel, E., Eds.; Pergamon Press: Oxford, 1982; Vol. 1, pp 555 et seq.
0 1983 American Chemical Society
Reactions of [(C,.le&02C1 J ( M = Rh or I r ) w i t h A12Me6
Organometallics, Vol. 2, No. 12, 1983 1725
N
c LL
r
t
h
a??
t-c
h
*m
wg
W
= * 0
1726 Organometallics, Vol. 2, No. 12, 1983
A
Vcizquez de Miguel et al.
i
Figure 1. 13C NMR spectrum of the solution obtained from reaction of [(C5Me,Rh),C14]with AlzMesin toluene-d, at -90 O C : (A) at 25 MHz, (B) at 100 MHz. Key: (a) C7D, solvent, (b) C5Me5, (c) C&fe5, (d) RhMe, (e) AlzMe6,(0 A12Me4C1,,(g) "lone" MeAl, (h) RhMe?.
Figure 2. 13C NMR spectrum of the solution obtained from reaction of [(C5Me51r),C14] with Al,Me6 in 2-methylbutaneat -120 O C and 25 MHz: (A) entire spectrum, (B) Al-methyl region. Key: (a) toluene reference, (b) C5Mea,(c) 2-methylbutanesolvent, (d) C&fe5, (e) AI,Me6, (f) AI2Me4Cl2,(g) "lonen methyl, Me-AI, (h)
IrMe,
experiments indicate that the rhodium-bound methyls at 6 -17.98 and +2.61 exchange only with each other (at rather higher temperature, broadening becomes marked only above ca. -30 "C), and the aluminum-bound methyls only exchange with each other. Exchange is slow, at least at the lower temperatures, between Rh- and Al-bound methyls. The complex is reasonably stable to ca. +40 "C, by which temperature all the methyl resonances are broadened, and the low-temperature spectra can be reproduced exactly on recooling. Further evidence concerning the nature of this species 3a comes from trapping experiments (Scheme I). A variety of ligands was added to these solutions at low temperatures, and the NMR spectra were examined. Ether ligands caused collapse and fast exchange of all the methyl resonances even at -90 "C (but see below), and amines and hexamethylphosphoramide had similar effects. Reaction with triphenylphosphine gave [CbMebRhMe2(PPh3)] (4a) in 49% yield, but solubility problems precluded NMR monitoring of the reaction. The best ligand for this purpose turned out to be dimethyl sulfoxide (Me,SO) since the reaction could be monitored at low temperature and a stable soluble complex, [C5Me5RhMe2(Me2SO)](5a), could then be isolated from the reaction in 95% yield. The properties of these and related complexes are described below. When Me2S0 was added to the yellow solution at very low temperature, only signals due to 3 and to the MezSO complex 5a could be detected in the NMR spectrum. Addition of further MezSO converted the spectrum into that of 5a together with that of an (uncharacterized) aluminum-MezSO complex, possibly A1MezC1(Me2SO). These experiments suggest that the rhodium in the yellow solution is still in the +3 oxidation state and that no reduction (or oxidation) has yet occurred. It should also be noted that, when the reaction of &Me6 with complex la is carried out very carefully at low temperatures (to give the yellow solution), no methane is liberated. Further, the 13CNMR spectra of the yellow solution show absolutely no sign of low-field (high-frequency) signals such as bridging or terminal CH or CH2ligands give
u) IrMe,.
rise to. This result may be contrasted with the formation of methane and of the trinuclear bis(y-methyne) cluster [ (C5Me5Rh)3(p-CH)z]7 which occurs when the same reactants are mixed at ambient temperature and in very concentrated solution. This reaction at low temperature may also be contrasted with the reaction to give 7 described by Tebbe et ala8
I 1/2Me,A12C12
Reaction of dichlorobis(p-chloro)bis(pentamethylcyclopentadieny1)diiridium (1b) with AlzMe6followed a quite parallel path at low temperature. Again a solution (very pale yellow-orange, in this case) was readily formed when a suspension of l b was reacted in a hydrocarbon solvent. Since the species present in this solution reacted slowly with aromatic compounds (toluene, benzene3),the solvents of choice, especially for the low-temperature 13C NMR experiments, were aliphatic hydrocarbons such as 2methylbutane or cyclopentane. (Parallel studies on the rhodium systems showed that the same species were formed in cyclopentane as in toluene.) The I3C NMR spectra at -90 "C (Figure 2) showed resonances at 6 -29.0 (2) and -15.7 (1)(relative intensities of the methyl resonances in parentheses), which we assign to the IrMe3, at 6 8.23 (5) and 93.2, assigned to the single C5MebIr,as well as methylaluminum resonances at 6 -3.9 and -8.0 (excess AlzMe6),-5.5 (8,AlzMe4Cl,),and -9.9 (0.51, assigned again to the lone methyl. These relative intensities were the same as those observed for the rhodium analogue. For this system the spectrum was also determined in 2-methylbutane at -120 "C. The chemical shifts (7) Vbquez de Miguel, A.; Isobe, K.; Bailey, P. M.; Meanwell, N. J.; Maitlis, P. M. Organometallics 1982, 1, 1604. (8) Tebbe, F. N.; Parshall, G. W.; Reddy, G. S. J. Am. Chem. SOC. 1978, 100, 3611.
Organometallics, Vol. 2, No. 12, 1983 1727
Reactions of [(CJ4e5M)2C1J ( M = R h or I r ) with A12Me6 of all the resonances were within 0.5 ppm of the values in cyclopentane a t -90 "C, and thus the low temperature limiting spectrum appears to have been reached, a t least so far as gross changes are concerned. Again, only a single iridium-containing species, [ (C5Me51rMe3),(A1Me)](3b), is present. When diethyl ether was added and the solution recooled to -118 "C, the 13CNMR spectrum showed resonances due to 3b virtually unshifted and still clearly recognizable as such. However the other methylaluminum resonances had all coalesced into a single sharp peak a t 6 -6.8. The fact that the peak at 6 -9.9, assigned to the "lone" methyl, is still clearly visible and not exchanging with the A12Me6and A12Me4C1,resonances is highly significant in the assignment of a structure to 3. Addition of Me2S0 simply and cleanly produced the [C5Me51rMez(Me,SO)]complex (5b) in high yield. When toluene or benzene was added to solutions of 3b, new resonances very soon appeared, consistent with our observation that these molecules reacted with the intermediate.3 T o sum up, both the rhodium and the iridium chloro complexes la and lb, respectively, react easily a t low temperatures with Al,Me6 to give a stoichiometric amount of AlzMe4Cl2and a thermally relatively stable species that is then defined as [(C5Me5MMeJ2A1Me] (3) produced according to 2[(C5Me5M),C14]+ 5A1,Me6 = 2[ (C5Me5MMe3),AlMe]+ 4A12Me4C12 3 One of the chief reasons for proposing this formula rather than the one suggested before, (A),lb is that the 100-MHz 13CNMR spectra strongly indicate that the intensity of the "lone" methyl is only half of that of the smaller of the Rh- (or Ir-) bound methyls. This points to a structure in which two Rh (or Ir) atoms and one A1 are in the unit. Further experiments were carried out with the aim of generating 3a from a different starting material. This was accomplished when a solution of the MezSO complex 5a was reacted with AlzMe6. The reaction proceeded quite slowly and required a large excess of &Me6 and gentle warming. In C7D8a t -90 "C signals were seen at 6 -0.6 (d, J = 28 Hz), 8.23 (s), and 98.4 (d, J = 3 Hz) due to complex 5a, a t 6 -5.8 which is where the dmso adduct of AlMe, resonates, a t 6 -4.7 and -6.9 (assigned to A1,Me6), and a t 6 -19.0 (d, J = 28 Hz), 1.6 (d, J = 23 Hz), 7.6, and 97.1. These last four resonances are very close to four of the five resonances found for 3a and we therefore assign them accordingly; the missing signal of the lone Al-bound methyl is probably under the A1Me,(Me2SO) resonance a t 6 -5.8. This experiment showed that complex 3a could also be generated by a different route. Structure of Complex 3. It is most unlikely that 3, or indeed any other species formed under these conditions, has any appreciable ionic character in view of the very high solubility in solvents such as 2-methylbutane a t temperatures as low as -120 "C. Indeed this observation strongly suggests that the molecules present must have a very "solvent-like" appearance-that is they must be relatively small and have an essentially all-hydrocarbon outer shell. The question then arises as to the disposition of the two types of rhodium- (or iridium-) bound methyls. One obvious solution to the problem is that they occupy three (two equivalent cis and one trans) sites in the base of a square pyramid. The four-legged piano stool geometry has been shown to occur in the Ir(V) complex [C5MeSIrMe4].3 However, since addition of MezSOand similar ligands gives
rise to well-defined M(II1) complexes, the fourth ligand would then have to be cationic in order to balance the charge, assuming the 18-electron rule continues to be obeyed. This implies the formation of a metal-metal bond (Al-Rh or Al-Ir; not Rh-Rh since that would result in a more complex spin-multiplicity for the attached methyls) and can be represented by structure 3-i. Me
Me
C~Me5\1
Me
ye
Me-M-AI-M-Me
II
II
Me
Me
Me
I AI
C5Me5.i
I,C~Me5
Me'
M ' e'
M 'e'
Me
I
M/'SMe5
M 'e
3-ii
3-i
One (indirect) analogy for structure 34 comes from the complexes [Cp,(C5H4)Mo2(p-H)z(p-AlMez)z(p-AlMe)] and [Cp,(C5H4),Moz(p-AlMez)(p-A1Me)] that both have two metal atoms held together by a bridging AlMe ligand.gJO However, in these molecules the aluminums also bind to the C5 rings, a feature for which the NMR spectra of 3 provide no evidence. A closer analogy is provided by the trimetallic complex 8 which arises from the reaction of [CpMo(CO),H] and GaMe,." [CpMo(CO),H] + GaMe, = [CpMo(CO),-GaMe2] +CH4 6[CpMo(CO),-GaMez] =
~[C~MO(CO),-G~M~-MO(CO)~C~] + 3GaMe3
8 For structure 3-i the lower intensity of the Rh- (or Ir-) bound methyl resonance would be due to the methyl trans to the Al, while the RhMez (or IrMeJ resonance would arise from the two equivalent methyls cis to the Al. Alternatively, either the pair of (equivalent) methyls or the single metal methyl could be bridging to the aluminum. Since the coupling to rhodium of the lower intensity doublet in 3a is smaller (23.3 Hz) than that of the larger doublet (28.1 Hz), which in turn is similar to that of the terminal methyls in 4a (29.5 Hz) or 5a (29 Hz), this suggests that the Me, ligands in 3a are also terminal and hence that it is the single methyl which bridges to aluminum as in structure 3-ii. This assignment is also consistent with the assignments for Al,Me6, where the bridging methyls resonate at lower field (higher frequency) than the terminal ones,12 and also for such complexes as [Cp,Sc(p-Me,)AlMe,] (9a), where again the methyls bridging Sc and A1 resonate a t substantially lower field (6 20.7) than the terminal methyls (6 -6.3).13 CH
Cp M'
3>AlMe2
C ' H,
9 a , M = Sc
b, M = Yb c,M=Ti d, M = Lu
One immediately obvious point of difference between 3-ii and the lanthanide complexes is that in the latter the metals are bridged by two methyls; this has been confirmed ~
~~
(9) Forder, R.; Prout, C. K. Acta Crystallogr., Sect. B 1974, B30,2312. (10)Rettig, S.J.; Storr, A.; Thomas, B. S.;Trotter, J. Acta Crystallogr., Sect. B 1974, B30,666. (11)(a) St. Denis, J. N.; Butler, W.; Glick, M. D.; Oliver, J. D. J . Organornet. Chern. 1977, 129, 1. (b) Conway, A. J.; Hitchcock, P. B.; Smith, J. D. J. Chem. SOC.. Dalton Trans. 1975. 1945. (12) Muller, N. Chem. Phys. 1962,36,359.Olah, G. A.; Prakash, G. K. S.; Liange, G.; Henold, K. L.; Haigh, G. B. R o c . N a t l . Acad. Sci. U.S.A. 1977. 74. 5217. (13)Holton, J. A.;Lappert, M. F.; Ballard, D. G. H.; Pearce, R.; Atwood, J. L.; Hunter, W. E. J.Chern. Soc., Dalton Trans. 1979,45 and 54.
Vizquez de Miguel et al.
1728 Organometallics, Vol. 2, No. 12, 1983 Scheme I1
Me,’
Me,,
CHMe,
I
Me,
by an X-ray structure determination of the Yb analogue 9b.13 A double bridge has also been proposed for [CpzTi(p-Me)2A1Me2](9c),13 [ (C5Me5)zLu(p-Me)zAlMez] (sa),l4 and [(q3-allyl)Ni (p-Me),A1Me2].I5 Structures 3-i and 3-ii formally leave the A1 in the trimetallic unit three-coordinate. This can be justified because of the bulk of the attached substituents; it is also possible that in solution there is some weaker interaction either between two units or between one and A1,Me4C12. The two representations 3 4 and 3-ii are to some extent equivalent since in molecules such as A12Me6which contain bridging methyls, at least some Al-Al interaction must also exist;16this will also apply to heterotrimetallic systems with bridging methyls. Further, if the ground state of 3 is closer to the metal-metal bonded form 3-i, then the reactivity of the molecule, in particular that with Me2S0 a t low temperature, suggests that the binding is quite weak. Indeed, the fact that on addition of an ether to a solution of 3a all the methyl resonances collapse even a t -90 “C indicates that all the methyl-metal bonds are very labile. The observation that a t -30 OC the Rh-bound methyls exchange only with each other and not with the Al-bound methyls is easily understood on the basis of the structures proposed and a facile interconversion between them as shown (Scheme 11). We also attempted to measure the 27AlNMR spectra of such solutions; however, a t temperatures below ca. -50 OC the resonances were so broad as to be unrecognizable. Since the molecules showed fast-exchange behavior already below this temperature, such measurements seemed rather unproductive. This effect arises from the high quadrupole moment of 27Al which also made a more quantitative treatment of the 13C exchange rates of doubtful value. The Formation of [(C5Me&hMe),(p-CHz),1 (2). We have previously described the reaction of the yellow solution containing 3a with acetone to give complex 2.’ It was of considerable importance to determine whether the aluminum played a vital role in the conversion of 3 into 2 or whether this could be achieved simply by hydrogen transfer from a “C5Me5RhMez”species. We have shown elsewhere” that the Me2S0 ligand in complex 5 is quite labile and that these complexes can be used as convenient sources of “C5Me5MMe2”. In fact, when 5a was heated with acetone (and a variety of other hydrogen-acceptor molecules), no reaction took place and no formation of 2 could be observed. By contrast, when 3a had been formed by prior reaction of 5a with &Me6, complex 2 was obtained in good yield on addition of acetone. This indicates that the aluminum plays an essential role in the conversion. The low-temperature 13C NMR spectrum of the solution
/ \
Me Me-AI-Me Me,CH’
C5Me5b 4 2 a R h V C 5 M e 5
dRh-
bMe
Me
Me
0’ \Me \H
+
3CH,
+
“ANOH),”
+
Me2CH0H
cis-2
obtained on addition of a small amount of acetone to a solution of 3a in deuteriotoluene a t low temperature showed the presence of only 3a and 2; no intermediates were detectable by NMR spectroscopy. Similarly, when the reaction was carried out in the cavity of an ESR spectrometer, no radical species could be detected either. This indicates a very facile reaction sequence from 3a to 2, and a proposal is made in Scheme 111. It will be noted that this sequence leads to the formation of cis-2 which is indeed the observed initial product, since the subsequent isomerization to trans-2 is promoted by Lewis acidse2 The reaction of 3b with acetone also gives some of the iridium analogue of 2, presumably formed by a similar reaction; however this is complicated by the formation of a variety of other p r o d ~ c t s . ~ J ~ The formation of [ (C5Me5Rh),(p3-CH),l7and of its iridium analogue, together with methane from the reaction of (&Me6) with complex 1 in excess, can also be understood in terms of the reaction represented by 2[ (C6Me5MMe3)2A1Me]+ [ (C5Me5M)&14]= 2[(C5MeSM)3(p3-CH)z] + A12Me2C14+ CH4
Complexes [C5Me5M(Me)2L](L = Me2S0, PPh3,and dppm/2). The Me’s0 complexes 5a and 5b were obtained in very high yield by reaction of l a or l b with A12Me6 and Me’SO; the precise yields depended somewhat on the exact conditions used. For preparative purposes it was convenient to make [C5Me5MClz(MezS0)]in situ by reaction of 1 with Me,SO and to then react this with A12Me6in toluene for 5a or pentane for 5b. The complexes were characterized by microanalysis and spectroscopic methods. The presence of a strong band in the IR a t 1085 cm-’ in both is consistent with the presence of an S-bonded Me’s0 ligand.lg This was confirmed by a careful analysis of the ‘H NMR spectrum of 5a which showed a small coupling (0.68 Hz) of the Me’s0 methyls to rhodium, characteristic of a three-bond H-C-S-Rh coupling.” The triphenylphosphine complexes 4a and 4b were similarly characterized, in particular by their ‘H NMR
~~
(14) Watson, P. L.J . Am. Chem. SOC.1982, 104, 337. (15) Fischer, K.;Jonas, K.; Misbach, P.; Stabba, R.; Wilke, G. Angew. Chem., Int. E d . Engl. 1973, 12, 943. (16) O’Neill, M. E.; Wade, K. ‘Comprehensive Organometallic Chemistry”; Wilkinson, G., Stone, F. G . A,, Abel, E., Eds.; Pergamon Press: Oxford, 1982; Vol. 1, p 12. (17) Gbmez, M.; Robinson, D. J.; Maitlis, P. M. J . Chem. SOC.,Chem. Commun. 1983, 825.
(18) Isobe, K.;Vizquez de Miguel, A,; Maitlis, P. M., unpublished results. (19) (a) White, C.;Thompson, S. J.; Maitlis, P.M. J. Chem. Soc., Dalton Trans. 1977, 1654. (b) Reynolds, W. L. Prog. Inorg. Chem. 1970, 12, 10. (c) Cotton, F. A.; Francis, R.; Horrocks, W. J . Phys. Chem. 1960, 64, 1534. (20) Barnes, J. R.; Goggin, P.L.; Goodfellow, R. J. J . Chem. Res., Synop. 1979, 118.
*
Reactions of [(C&fe5M)2Cl J ( M = R h or Ir) with A12Me6
m c
M
Figure 3. Manifold used for loading NMR tubes (d) with A12Me6 solutions under argon: (a) RB-19, (b) stopcocks, (c) RB-10 joint.
spectra which showed the iridium-bonded methyls, in 4b, as a doublet [J(P-H) = 5.3 Hz], and the rhodium-bonded methyls, in 4a, as a double doublet [J(P-H) = 4.7 Hz, J(Rh-H) = 2.3 Hz]. Complexes 4a and 4b were also conveniently formed by reaction of 5a and 5b respectively, with triphenylphosphine in refluxing benzene. The bis(dipheny1phosphino)methaneligand reacted with either 3a or the MezSO complex 5a to give the bridged dinuclear complex 6. The proton NMR spectrum was normal (Table I), but the 31PNMR showed two "triplets" which arose from an [AX], spin system owing to the magnetically inequivalent phosphorus nuclei. This could be analyzed in a conventional mannerz1 to give 2J(P-P) = 45.2, 'J(Rh-P) = 171.9, and 3J(Rh-P) = 4.3 Hz a t a chemical shift of 6 45.4 with respect to external phosphoric acid.
Experimental Section All manipulations involving the use of organoaluminum compounds were carried out under a protective atmosphere of argon in carefully dried equipment. Standard Schlenk apparatus was used and liquids were transferred by syringe or cannulae. Samples for NMR measurements (5-mm tubes for 'H, 10-mm tubes for I3C) were made up under argon by using a simple piece of equipment as shown in Figure 3. NMR spectra were measured at 60 (R-12B) and 400 MHz (WH-400) for 'H and at 25 (PFT-100) and 100 MHz (WH-400) for 13C. The I3C spectra measured in hydrocarbon solvents were referenced against the toluene resonance at 6 21.34 as internal standard; other spectra were referenced against tetramethylsilane (6 0) or dichloromethane (6 5.33 for 'H) as appropriate. IR spectra were measured on a P.E. 157 or on a P.E. 180 with a far-IR attachment. Microanalyses were carried out by the University of Sheffield Microanalytical Unit. Analytical and spectroscopic data for all the new complexes isolated are given in Table I. Preparation of Complex 3a. Complex la (0.24 g, 0.38 mmol) was transferred into a 10-mm diameter NMR tube (d, Figure 3; usually of quartz, for extra strength) fitted with a RB-10 joint (c) attached to the manifold containing two stopcocks (b) and a RB-19joint (a) covered with a Suba-seal (Figure 3). The tube containing la was heated to 80 "C and alternately evacuated and filled with pure argon some 10-12 times during 1 h. It was then cooled to 20 "C, solvent (deuteriotoluene, cyclopentane, etc. Caution: Solutions of A12Me6 in low boiling solvents are unusually pyrophoric and dangerous) added, the mixture injected onto the solid sample and then cooled further a t -78 "C under a flow of argon. A solution of A1,Me6 (usually 10 % w/w in the (21) Harris, R. K. Can. J . Chern. 1964, 42, 2275.
Organometallics, Vol. 2, No. 12, 1983 1729 same solvent, 2 5 1 molar ratio) was then added dropwise over 10 min. The solid dissolved to give a pale yellow solution that was briefly shaken a t -78 "C to complete the reaction. The manifold was removed under a steady stream of argon and quickly replaced by a Suba seal to cap the NMR tube. With only a little practice it is possible to prepare in this manner clear solutions containing A12Me6that show no trace of the A13Me6(OMe)22 resonances a t 6 -9.3 and 51.6 (in toluene) and that are quite free of contamination by moisture or oxygen. A t -90 "C in deuteriotoluene (reference 6 21.32) the solution showed resonances at 6 97.89 (d, J = 2.7 Hz), 8.52, 2.61 (d, J = 23.3 Hz), -4.54, -5.03, -5.89, -7.19, and -17.98 (d, J = 28.1 Hz). The relative intensities and the assignments of the peaks are discussed in the text. When the same solution was made up in cyclopentane (containing a small amount of toluene as reference), the same resonances appeared a t 6 97.87, 8.72, 2.50 (d, J = 22.9 Hz), -4.09, -5.42, -6.94, -8.03, and -17.92 (d, J = 27.5 Hz). For comparison the resonances of pure A12Me6appeared at 6 -4.60 and -7.15 (toluene) and at 6 -4.09 and -7.73 (cyclopentane) under the same conditions. When a small amount of deuterioacetone was added to a solution of 3a in deuteriotoluene, all the resonances decreased and two new resonances, due to the cis isomer of complex 2, appeared at 6 9.7 and 100.0; no other resonances were observable. Similarly, when MezSO was added to a solution of 3a a t low temperatures, only the resonances due to 5a and to 3a could be observed. When a small amount of tetrahydrofuran was added to a solution of 3a in deuteriotoluene, the spectra a t -80 'C (and at -50 "C) showed resonances due to two CSMe,Rh species, at 6 98.85 (d, J = 3.8 Hz) and 9.45 and at 6 99.35 (d, J = 3 Hz) and 10.85. However all the methyl resonances had collapsed to a singlet a t 6 -8.55. A solution of the rhodium-Me2S0 complex 5a was made up in deuteriotoluene, and a large (7-8 fold) excess of A12Me6was added; the solution was then warmed to +40 "C before cooling to -90 "C to measure the I3C spectrum. This showed resonances at b 98.35 (d, J = 3.1 Hz), 8.23, and -0.61 (d, J = 28.2 Hz) due to unreacted 5a, at 6 -4.75 and -6.85 due to AlzMe6,and at 6 -5.85 due to Al(Me2SO)Me3,as well as the following peaks assigned to 3a: 6 97.1 (d, J = 3.8 Hz), 7.62, 1.57 (d, J = 23.7 Hz), -8.24, and -19.0 (d, J = 28.2 Hz). Preparation of Complex 3b. This was prepared, in cyclopentane containing a small amount of toluene as reference, in exactly the manner described for the preparation of 3a. At -90 "C this showed the following resonances: 6 93.23, 8.23, -9.91, -15.73, and -29.02 due to 3b, 6 -3.90 and -7.97 due to A12Me6, and at -5.48 due to A12Me4C12.The NMR spectrum in 2methylbutane a t -120 "C was similar and showed resonances at 6 93.74, 8.99, -8.51, -14.98, and -28.32 due to 3b as well as at 6 -3.51, -4.42, and -6.69 due to the organoaluminum species. On addition of diethyl ether and remeasuring the spectrum a t -118 "C the resonances due to 3b remained (at 6 93.62, 8.72, -9.03, -15.65, and -28.75) but the methylaluminum resonances had all coalesced to a singlet a t 6 -6.76. Preparation of the Iridium-Me,SO Complex 5b. Method A. Complex l b (0.50 g, 0.63 mmol) and a small magnetic stirrer were placed into a dry Schlenk tube and heated (5 min/60 "C); the tube was then filled with dry argon. This procedure was repeated six times over 30 min. The tube was then cooled to 20 "C and dry n-pentane (25 cm3) added to the solid; the suspension was stirred and cooled to -78 "C. A freshly prepared solution of hexamethyldialuminum in n-pentane (2.6 cm3 of a 26.5% w/w solution; 6.3 mmol) was added dropwise with stirring at -78 "C. The orange suspension gradually dissolved to give a pale yellow-orange solution. After the solution was stirred (10 min a t -78 "C), excess dimethyl sulfoxide (0.3 cm3) was added dropwise over 20 min; the solution was allowed to warm to 20 "C and all volatiles were removed in vacuo. The yellow-orange residue was extracted with wet pentane (3 X 40 cm3);this solution was filtered and evaporated to dryness, and the extraction was repeated to give a pentane solution of complex 5b which deposited yellow crystals of the complex (0.27 g, 50%) on cooling. Method B. A suspension of complex lb (0.50 g) in n-pentane (25 cm3) was prepared as described above. T o this was added (22) Sakurada, Y.; Huggins, M. L.; Anderson, W. R., Jr. J . Chem. Phys. 1964, 68, 1934 and references therein.
1730
Organometallics 1983, 2, 1730-1736
Me2S0 (0.2 cm3, 2.8 mmol); after the solution was stirred for a few minutes, the suspension was cooled to -78 O C and hexamethyldialuminum (2.1 cm3of a 26.5% w/w solution in pentane, 5.0 mmol) was added dropwise. The suspension slowly turned yellow and after 30 min was allowed to warm to 20 O C . This solution was worked up as described above to give yellow crystals of 5b (0.47 g, 80%). This complex could also be made from [C5Me51rC12(Me2SO)] if this was isolated from the reaction of Me2S0 with complex lb. The rhodium complex 5a as well as the complexes 4a, 4b, and 6 was all prepared by method A and are detailed in Table I. In addition complexes 4a, 6,and also 4b were prepared by reaction of the MezSO complexes 5a and 5b, respectively, with the appropriate ligand (method C). Preparation of Complex 6. Method C. A solution of bis(dipheny1phosphino)methane (0.11g, 0.29 mmol) was added to a solution of complex 5a (0.10 g, 0.29 mmol) in dry benzene (15
cm3). The solution was refluxed (1h), the solvent removed, and the residue extracted with wet pentane. On cooling, the pentane solution slowly gave yellow crystals of complex 6 (0.20 g, 75%).
Acknowledgment. We thank the S.E.R.C. for supporting this work, the Spanish Ministry of Education of the award of a scholarship (to A.V.deM.), Professor N. M. Atherton for ESR spectra, Ethyl Corp. for organoaluminum compounds, Dr. W. Schaefer (Chemischewerke Huels) for organic starting materials, and Johnson Matthey for the loan of iridium chloride. Ftegistry No. la, 12354-85-7;lb, 12354-84-6;3a, 87174-69-4; 3b, 87174-70-7;4a, 87174-71-8;4b, 87174-72-9;5a, 87183-26-4;5b, 87183-27-5;6a, 87174-73-0;A12Me6,15632-54-9;Ph3P,603-35-0; Me2S0, 67-68-5; A12Me,C12, 14281-95-9;dppm, 2071-20-7; Al, 7429-90-5; Rh, 7440-16-6;Ir, 7439-88-5; acetone, 67-64-1.
Kinetics and Mechanism of the Phase-Transfer-Catalyzed Carbonylation of Benzyl Bromide by the Cobalt Tetracarbonyl Anion Hew6 des Abbayes,' Anne Buloup, and Guy Tanguy Laboratoire de Chimie des Organom&alliques, ERA CNRS No. 477, Universit6 de Rennes, Campus de Beaulieu, 35042 Rennes Cedex, France Received March 16, 1983
The carbonylation of benzyl bromide with the cobalt carbonyl anion according to eq 1and 2 was studied in a liquid-liquid two-phase system (H20/CH2C12,H20/i-Pr20,H20/C,HG). The disappearance of benzyl bromide was found to follow a second-order kinetic law under stoichiometric conditions. In these experiments, the concentration of the cobalt carbonyl anion in the organic phase remained constant throughout the reaction, owing to rapid pairing with the tetrabutylammonium cation used as a cationic carrier. Under catalytic conditions, the kinetics depended strongly on the stirring speed (which was not the case under stoichiometric conditions). A catalytic cycle is proposed for this liquid-liquid system, in which it is shown that the catalyst (Bu,N+CO(CO)~-) stays in the organic phase, and the product (the phenylacetate anion) is expelled into the aqueous phase. It is also shown that the rate-limiting reaction is not the nucleophilic displacement (1)(reaction of benzyl bromide with the cobalt carbonyl anion) but the cleavage of the acylcobalt carbonyl 4 by the ion pair Bu4N+OH-at the liquid-liquid interface. The presence of byproducts such as toluene and diphenylethane is accounted for in terms of the decomposition of a transient hydroxycarbonyl species 13 through loss of carbon dioxide. One of the main challenges in the field of catalysis is to combine the advantages of heterogeneous and homogeneous catalysis. T h e former (mostly solid-gas) is very efficient, securing a good separation between the catalyst and the product, but lacks selectivity. The latter proceeds under mild conditions and is more selective, but catalyst and products are mixed in the same liquid phase. Several years ago, Bailar' suggested a liquid-liquid biphasic system which, in principle, should secure separation between the catalyst and the products by a proper choice of two immiscible liquids A and B. Useful combinations of catalyst, reagents, and products in such a liquid-liquid two-phase system are given in Scheme 1. Recent applications of phase-transfer catalysis2 to organotransition-metal reactions3 represent a step in this (1)Bailar, J. C. Catal. Reu. 1974 10, 17. (2) (a) Weber, W. P.; Gokel, G. W. "Phase Transfer-Catalysis in Organic Synthesis"; Springer Verlag: Berlin, Heidelberg, New York, 1977. (b)Starks, C. M.; Liotta, C. L. 'Phase-Transfer Catalysis, Principles and Techniques"; Academic Press: New York, 1978. ( c ) Montanari, F.; Landini, D.; Rolla, F. Top. Curr. Chem. 1982,101, 147-200.
0276-7333/83/2302-1730$01.50/0
Scheme I liquid A catalyst liquid B reagents
+ products liquid A catalyst + reagents liquid B products
direction. Expected effects are as follows: (1)good separation between the catalyst and the product; (2) increased efficiency; (3) new selectivity due to the unusual conditions met within two-phase systems. A number of experimental results have already been r e ~ i e w e dbut, , ~ so far, no mechanistic investigation has been attempted. We report here a study of the catalytic carbonylation of benzyl bromide, run in a liquid-liquid phase-transfer system, with the cobalt carbonyl anion as (3) des Abbayes, H.; Alper, H. J . Am. Chem. SOC.1977,99,98. (4) (a) Cassar, L. Ann. N . Y. Acad. Sci. 1980,333,208.(b) Alper, H. Adu. Organomet. Chem. 1981,19, 183.
0 1983 American Chemical Society
